The Beveridge curve still indicates low probability for a soft landing

By: Alex Domash¹ and Lawrence H. Summers²

In a recent paper published with Olivier Blanchard in July 2022, we argued that the Fed’s hope to cool the labor market through a decrease in job vacancies alone — without an increase in unemployment — was highly implausible. The empirical evidence shows that there has never been a historical example where job vacancies came down in a substantial way without a significant increase in unemployment.

Over the last few months, however, some have questioned our conclusions on the basis that job vacancies have started to fall from their March 2022 peak, with little movement in the unemployment rate. Between March 2022 and July 2022, the total number of job openings declined from 11.8 million to 11.2 million. Meanwhile, the unemployment rate has fallen from 3.6 percent to 3.5 percent.

Rather than contradicting our conclusions, the recent movement in job vacancies continues to show very little evidence for a soft landing for two reasons:

1. Job openings in July 2022 remain near record highs and show little sign of cooling. While data from last month’s Job Openings and Labor Turnover Survey (JOLTS) showed a large drop in job vacancies for June, recently released data included a large upward revision for June job openings, and an increase in job openings in July to 11.2 million. As such, the labor market still has a long way to go before wage pressures begin to ease.
2. Even if job vacancies had declined, vacancies always decline significantly months before the unemployment rate begins to increase. On average, when job vacancies decline by 10 percent from their peak, the unemployment rate contemporaneously increases by only 0.2 percentage points. However, in the 12 months after vacancies fall 10 percent from their peak, the unemployment rate increases, on average, by 2.5 percentage points. This suggests that an initial fall in the job vacancy rate, with no contemporaneous movement in the unemployment rate, is to be expected based on the historical experience.

Given these two observations, we continue to believe that fighting inflation will require a reduction in job vacancies and also an increase in unemployment.

Figure 1 shows the Beveridge curve (the relationship between job openings and unemployment) with the recently released JOLTS data, which includes job vacancy data through July. The outward shift in the Beveridge curve continues to be pronounced, with very low unemployment and very high job vacancies. In July, the job vacancy rate was 6.9% (up from 6.8% in June), while the vacancy to unemployment ratio was at a record high at 2.0 (up from 1.9 in June).

In our July 2022 paper, we looked at the historical relation between job vacancies and unemployment going back to the 1950s and showed that there has never been a historical example where the job vacancy rate came down in a substantial way without a significant increase in unemployment. We also discussed the existence of “Beveridge loops” — counterclockwise movements in the vacancy/unemployment space, which arise since firms can adjust vacancies instantaneously, while unemployment adjusts only as layoffs and hires take place. This implies that vacancies usually decrease after a peak before the unemployment rate starts to increase.

Figure 2 uses historical data to demonstrate the existence of Beveridge loops. The figure plots separately each previous vacancy rate peak (red), along with i) the movement in unemployment when vacancies fall 10 percent from a peak (green), and ii) the increase in unemployment within the subsequent 12 months after vacancies decline (blue). The figure shows that a 10 percent decline in the vacancy rate from a peak usually occurs with very little change in the unemployment rate (movement from the red to green dots). However, in the 12 months after the vacancy rate declines by 10 percent, the unemployment rate always increases substantially (movement from the green to blue dots).

To the extent that job vacancies have declined in recent months, this evidence suggests that a decline in job openings — with no contemporaneous increase in the unemployment rate — is fairly typical, and offers little reason for optimism about the chances of a soft landing.

Table 1 presents the same evidence in quantitative terms. For each episode, it computes the change in the unemployment rate from the month of the peak vacancy rate to i) the month when vacancies fall 10 percent from a peak, and ii) the month when unemployment reaches a maximum over the subsequent 12 months.

The evidence suggests that the vacancy rate almost always falls substantially before the unemployment rate increases: historically, a 10 percent decline in the vacancy rate is associated with an average contemporaneous increase in the unemployment rate of only 0.2 percentage points. But the data also shows that the unemployment rate always increases substantially in the 12 months following a steep decline in vacancies: on average, a 10 percent fall in vacancies is associated with a 2.5 percentage point increase in the unemployment rate over the subsequent year.

Some have critiqued the use of job openings data for evaluating labor market tightness, arguing that data on openings may be overstated since firms may not imminently intend to fill all job listings. But the nature of a JOLTS vacancy has not changed over the last two years, and requires that “there is active recruiting for workers from outside the establishment location” and that “the job could start within 30 days.” We have also previously shown that job vacancies are comparable to unemployment in predicting wage inflation, and that there has not been much of a trend in peak vacancies going back to the 1950s. Finally, the job quits rate tends to covary with job vacancies, and quits have exhibited a very similar pattern to job vacancies in recent months (total quits have decreased by 6% since March, compared to a 5% decline for vacancies). For these reasons, we believe the vacancy data is a credible signal of labor market tightness.

Overall, the data suggests that job vacancies still have a long way to go before they are at a level compatible with the Fed’s inflation target. And while the unemployment rate has remained at historic lows, we show that the empirical evidence implies that job vacancies almost always fall significantly before unemployment starts to increase. The information from the labor market indicates that it is far too soon for a victory lap regarding the prospects of a soft landing for the US economy.

[1] Alex Domash is a Research Fellow in the Mossavar-Rahmani Center for Business & Government at the Harvard Kennedy School. He is recent graduate of the Masters in Public Administration in International Development (MPA/ID) program at the Harvard Kennedy School. Twitter: @asdomash

[2] Lawrence Summers is the Charles W. Eliot University Professor and President Emeritus at Harvard University. He served as the 71st Secretary of the Treasury for President Clinton and the Director of the National Economic Council for President Obama. Twitter: @LHSummers